Vector Auto Regressive (VAR) Analysis

The VAR analysis has been performed to see the effect of policy changes on banks behavior and to examine whether that changes induce banks to do speculative behavior during the bubble period. Semi-annual data from 1977-2004 has been used for the domestically licensed banks.

Unrestricted VAR analysis has been performed to see the impulse responses of variables11.

The variables that are considered: Banks credit growth (LOANGR), Growth of investment and securities of banks (INVGR), Consumer price index (CPI, 1990=100), Growth of money Supply (MONEYSUPPLY), BoJ’s Discount rate (DISC_RATE), Growth rate of GDP (GDP), Growth of interest rate (GINT), Agency Conflict (ACONFLICT), Moral Hazard (MORAL_HAZ), Real Effective Exchange Rate Appreciation (EX_Appreciation), and Banks Profitability (Y) defined in section 5.1.

However, I construct several panels of VAR following Sims (1980, 1986, 1992) since VAR approach sometimes quiet helpful in examining the relationships among a set of economic variables although there are many criticisms on VAR. Consider

Vyt= Bi(Liyt) + et

where yt denote the vector of endogenous variables, the matrices V and Bi being conformable

with y-vector; Liyt denote the ith-lagged y-variable. The VAR is estimated with its reduced form

as

yt = V-1Bi(Liyt) + εt ; εt =V-1et.

From the estimated VAR, I study the ‘innovations accounting’ that is the impulse responses of different variables on one standard deviation for examining structural system.

As the ordering of the variables is important in VAR analysis, following Sims (1992) and empirical analysis of Ford et al. (2003), I placed the policy variables (e.g. monetary policy

11

Although there is subtle evidence of cointigration but I found that the implied order of Vector Error Correction (VEC) model is not the appropriate method here from which we may extract impulse responses. As Canova (1995) mentioned that even when the data are non-stationary there is no requirement to transform the VAR into VEC for meaningful inferences.

variables) at first and than the non-policy variables such as GDP, CPI, LOANGR etc. If the

correlations between the variables are negligible, that is if |ρij| <0.2, then ordering is immaterial.

But this is not the case in this study.

The Phillips-Perron test and the Breusch-Godfrey LM tests have been used to test serial correlation. Dickey-Fullar test has been used for testing unit root process. Since most of the variables are considered as growth rate, these follows I(0) processes and a few render I(1) process that is the first difference is necessary to become stationary.

Figure 5 displays the impulse responses of bank credit growth to different policy shocks such as growth of money supply, discount rate, growth of interest rate and real effective exchange rate appreciation. Solid lines represent point estimates while dashed-lines denote plus-2 and minus-2 standard deviation innovations. It is seen that money supply growth and exchange rate had negligible impact on banks credit extensions. But discount rate and interest rate show increasing effect on bank credit extensions.

Figure 6 displays impulse response of CPI, ROA, GDP growth and investment growth to bank credit shocks. All these variables respond strongly to bank credit expansions. This is consistent with the view that bank expanded their credits to real estate business and SME market in the 1980s due to pressure of low profit. Moreover, it is also consistent that changes in investment and securities were caused by changes in bank credit, where deregulations came into effect with the access of banks to short-term bond market in the late 1980s.

Figure 7 displays impulse response of bank low profitability (Y) to money supply, discount rate, increase of interest rate and exchange rate appreciation. Although growth of money supply shows some cyclical variations, other variables remain flat to the low profitability of banks. Therefore, it is interesting to make comments that monetary policy had negligible impact on banks low profitability. This finding is consistent with Hossain’s (2005) finding that banks low profitability is mainly generated from weak corporate governance and other problems. So it is

important to investigate the impact of agency conflict and moral hazard issues that arise from the policy inconsistencies.

Figure 8 displays that impulse response of bank low profitability (Y) to agency conflict and moral hazard is negative. This is consistent with our proposition that these two issues created room for banks to make speculation to increase profit during the 1980s. Therefore as it is expected, these two variables showed negative variations with low profitability.

Figure 9 displays strong and positive impulse response of credit growth to agency conflict and moral hazard. This is also consistent with the argument that moral hazard and agency conflict help bank to expand credit aggressively in the 1980s, which ultimately made banks vulnerable to crisis. In one hand, due to lack of prudential regulations, ongoing financial deregulations created moral hazard problem, and on the other hand, policy inconsistencies created conflict of interest among the respective agencies. Both the problems induced banks to behave aggressively which ultimately contributed to the prolonged banking crisis in Japan.

Figure-5: Impulse Response of bank credit growth to money supply, discount rate, growth of interest rate and exchange rate appreciation innovations

-1000 -500 0 500 1000 1500 2000 5 10 15 20 25 30 Response of LOANGR to M2CD -1000 -500 0 500 1000 1500 2000 5 10 15 20 25 30

Response of LOANGR to DISCRATE

-1000 -500 0 500 1000 1500 2000 5 10 15 20 25 30

Response of LOANGR to GINT

-1000 -500 0 500 1000 1500 2000 5 10 15 20 25 30

Response of LOANGR to EXGR Response to Cholesky One S.D. Innovations ± 2 S.E.

Figure-6: Impulse Response of CPI, ROA, GDP growth rate, growth of interest rate to banks credit extensions -40 -20 0 20 40 60 5 10 15 20 25 30

Response of CPI to LOANGR

-2 -1 0 1 2 3 4 5 10 15 20 25 30

Response of ROA to LOANGR

-30 -20 -10 0 10 20 30 40 50 5 10 15 20 25 30 Response of GDP to LOANGR -40 -20 0 20 40 60 80 5 10 15 20 25 30

Response of INVGR to LOANGR Response to Cholesky One S.D. Innovations ± 2 S.E.

Figure-7: Impulse Response of banks low profitability shocks to money supply, discount rate, growth of interest rate and exchange rate appreciation innovations

-3 -2 -1 0 1 2 3 5 10 15 20 25 30 Response of Y to M2CD -3 -2 -1 0 1 2 3 5 10 15 20 25 30 Response of Y to DISCRATE -3 -2 -1 0 1 2 3 5 10 15 20 25 30 Response of Y to GINT -3 -2 -1 0 1 2 3 5 10 15 20 25 30 Response of Y to EXGR Response to Cholesky One S.D. Innovations ± 2 S.E.